Developing a Virtual Flowability Sensor for Monitoring a Pharmaceutical Dry Granulation Line

Area of Science:

• Pharmaceutical Sciences
• Chemical Engineering
• Materials Science

Background:

• Current methods for measuring granule flowability are time-consuming at-line processes.
• Real-time monitoring is crucial for quality assurance in continuous dry granulation and tableting lines.
• A need exists for inline sensors to assess powder flowability immediately after roller compaction.

Purpose of the Study:

• To develop a real-time sensor for granule flowability using particle size and shape data.
• To evaluate the effectiveness of Partial Least Squares (PLS) regression in predicting flowability.
• To demonstrate the utility of particle analyzers as inline flowability sensors.

Main Methods:

• Utilized Partial Least Squares (PLS) regression to model granule flowability.
• Incorporated distributions of granule size and shape (eccentricity, form factor, elliptical form factor) into the PLS model.
• Validated predictions against three established flowability measurements: rotary drum flow, orifice flow, and tapped density analysis.

Main Results:

• Achieved high prediction performance with coefficients of determination ranging from 0.80 to 0.97.
• Demonstrated superior predictive accuracy compared to existing literature methods.
• Highlighted the effectiveness of PLS in handling multicollinearity and high-dimensional data.

Conclusions:

• Particle size and shape analysis, when combined with PLS regression, can reliably predict granule flowability in real-time.
• The study advocates for incorporating granule shape characteristics into flowability assessments.
• PLS regression offers an advanced approach for dimensionality reduction of particle distribution data, outperforming traditional percentile methods.